Synthesis of water-soluble surfactants using catalysed condensation polymerisation in green reaction media†

Sustainable and biobased surfactants are required for a wide range of everyday applications. Key drivers are cost, activity and efficiency of production. Polycondensation is an excellent route to build surfactant chains from bio-sourced monomers, but this typically requires high processing temperatures (≥200 °C) to remove the condensate and to lower viscosity of the polymer melt. In addition, high temperatures also increase the degree of branching and cause discolouration through the degradation of sensitive co-initiators and monomers. Here we report the synthesis of novel surface-active polymers from temperature sensitive renewable building blocks such as dicarboxylic acids, polyols (d-sorbitol) and fatty acids. We demonstrate that the products have the potential to be key components in renewable surfactant design, but only if the syntheses are optimised to ensure linear chains with hydrophilic character. The choice of catalyst is key to this control and we have assessed three different approaches. Additionally, we also demonstrate that use of supercritical carbon dioxide (scCO2) can dramatically improve conversion by reducing reaction viscosity, lowering reaction temperature, and driving condensate removal. We also evaluate the performance of the new biobased surfactants, focussing upon surface tension, and critical micelle concentration.

Where T: volume NaOH (aq) titrated to reach the equivalence point (mL), M: molarity of NaOH (aq), W: weight of sample (g), AVx: acid value at the start (t0) and end (t) of the reaction. As acid values are defined by the number of mg KOH equivalent to the acid content of one gram of sample, 1 the mass of KOH (56.1 g/mol) has been used for calculations.
The hydroxyl value (OHV) of the polyesters was determined by hydroxyl end-group titrations using aqueous NaOH (~ 0.6 M). It was determined to ensure that the synthesised polyesters retain their hydrophilic OH groups and to assess the ratio between OH:COOH groups for each compound. The technique is based on reacting a compound with free hydroxyl groups with a known excess of acetylating reagent (a solution of acetic anhydride in pyridine 5% (V/V)). Followed by hydrolysing the unreacted acetic anhydride to acetic acid by treating with deionized water and then titrating the resulting solution with NaOH. A typical OHV determination was performed as follows: 0.25 g of polyester were accurately weighed into a 250 mL RBF. Acetylating reagent was added to the RBF (5 mL) and a water condenser was attached to the flask neck. The solution was heated (to ~ 90-100 °C) using a heating block for 1 hour and stirred magnetically to allow the free hydroxyl groups to react with the acetic anhydride. After 1 hour; 10 mL deionised water was added through the top of the condenser and the mixture was heated for a further 10 minutes to hydrolyse the remaining unreacted acetic anhydride to acetic acid. The RBF was removed from the heating block and allowed to cool to RT. Once cool, the condenser was washed down with THF (90 mL). This solution was placed on the stirring mantel of the auto-titrator and titrated against a 0.6 M NaOH solution. The measurement was performed in duplicate, along with a blank which followed the exact same procedure with the exception that the sample was omitted. The OHV and the degree of polymerizations were then calculated from the resulting data using the following formulae: Where B: volume NaOH (aq) required for the blank (mL), T: volume NaOH (aq) required for the sample (mL), M: molarity of NaOH (aq), W: weight of sample (g), AV t : acid value at the sample (mg KOH per g). As the hydroxyl value is defined by the number of mg KOH equivalent to one gram of sample, 1 the mass of KOH (56.1 g/mol) has been used for calculations. When determining DP OHV the AV t has been multiplied by 4 as in theory there will be 4 hydroxyl groups for each repeating D-sorbitol unit along the polymer backbone. Table S1 below shows the data obtained from the potentiometric measurement of PSA with different catalysts.

Estimation of theoretical acid and hydroxyl values
A repeat unit of PSA has a molar mass of 287 g/mol. PSA of 1700 g/mol therefore has a degree of polymerization of 5.9. In an equimolar reaction of D-sorbitol and adipic acid, if the reaction would go to completion and only the primary alcohols react, there would be 4 free hydroxyl groups per repeat unit of PSA. We would also have one terminal hydroxyl and carboxylic acid group. That would be about 25 free hydroxyl groups and 1 free carboxyl group in the 1700 g/mol PSA example. The theoretical OH:COOH ratio for this polymer is therefore 25:1. Table S1 shows that the acid value (AV t ) is 34 while the hydroxyl value (OHV t ) is 677 which lead to an experimental OH:COOH ratio of about 20:1. In the same way, the theoretical OH:COOH value of PSA of 1100 g/mol synthesized with Novozym is , the value for PSA of 3300 g/mol is estimated to be 45:1 while that of PSS of 5900 g/mol is 87:1.  2. b Polymerization was carried out in bulk. Note that the OH:COOH ratio of the polyesters is likely a slight underestimation compared to the true value, as acetylation of all hydroxyl groups would be challenging due to steric hindrance. This underestimation is seen in both the K 2 CO 3 and the enzyme catalyzed samples.

Bulk polycondensation
Novozym 435 catalyst D-sorbitol (5.00 g, 27.5 mmol) and adipic acid (4.02 g, 27.5 mmol) were transferred into a 50 mL threeneck round bottom flask (RBF) equipped with a mechanical stirrer and argon sparge. The reagents were heated (120 o C, approx. 15 min) under mechanical stirring (300 RPM) until molten. Reaction temperature was lowered (95 o C) and Novozym 435 (10 wt% relative to monomers) was added. The reaction was terminated after 48 hours by allowing the reaction mixture to cool to ambient temperature. Once cooled the product was dissolved in a 1:1 (v:v) mixture of MeOH and H 2 O and filtered to remove the enzyme beads. The filtrate was recovered by solvent evaporation under reduced pressure and cooled (-18 o C) before being freeze-dried (Mini-Trap freeze dryer, LTE Scientific, UK).

Para-Toluenesulfonic acid catalyst
D-sorbitol (5.00 g, 27.5 mmol), adipic acid (2.01 g, 13.8 mmol), and pTSA (2 wt% relative to monomers) were added to a RBF. The mixture was heated (120 o C) and mechanically stirred (300 RPM). The reaction was terminated after 24 hours by decanting the reaction mixture into a ceramic crucible. The collected product was dried in vacuo (100 mbar, ambient temperature, overnight).

Potassium carbonate catalyst
D-sorbitol (5.00 g, 27.5 mmol), adipic acid (4.02 g 27.5 mmol), and potassium carbonate (5 wt% relative to monomers) were added to a RBF. The mixture was heated (120 o C) and mechanically stirred (300 RPM). The reaction was terminated after 48 hours by decanting the reaction mixture into a ceramic crucible. The collected product was dried in vacuo (100 mbar, ambient temperature, overnight).

Varying dicarboxylic acid chain length in base catalysed bulk polycondensation
D-sorbitol (20.00 g, 110 mmol), dicarboxylic acid (110 mmol, Table S2), and potassium carbonate (5 wt% relative to monomers) were transferred into a 100 mL three-neck RBF equipped with a mechanical stirrer and argon sparge. The mixture was heated (120 o C) and mechanically stirred (300 RPM). The reaction was terminated after 48 hours by decanting the reaction mixture into a ceramic crucible. The collected product was dried in vacuo (100 mbar, ambient temperature, overnight).

Polycondensation of D-sorbitol and natural diacids in supercritical CO 2
Reactions were performed in a 60 mL high pressure autoclave described previously. In a typical reaction, D-sorbitol (5.00 g, 27.5 mmol), adipic acid (4.02 g, 27.5 mmol), and potassium carbonate (5 wt% relative to monomers) was added to the base of the autoclave. The autoclave was degassed using a flow of scCO 2 (2 bar) for 15 minutes before being sealed and pressurised (55 bar) by addition of scCO 2 . The autoclave was subsequently heated (120 o C) before additional scCO 2 was added to raise the internal pressure to 240 bar. Consequently, the reaction was either allowed to proceed, or extracted during the last 15 minutes of every hour (6, 12, 24, 48 h). Products was collected by allowing the autoclave to cool to ambient conditions and depressurising slowly. The results of the synthesis for PSA and PSS are shown in Table S3 below.

Table S3
Reaction of D-sorbitol and natural diacids in a 1:1 ratio using scCO 2 with extraction at 120 °C.

End-capping of sorbitol-based polyesters with lauric and stearic acid
Reagents (Table S 2) and catalyst (K 2 CO 3 , 5 wt% w.r.t. fatty acid) were transferred into a 50 mL 3 neck RBF equipped with a mechanical stirrer. The mixture was degassed (15 min, argon) before being immersed in a pre-heated oil bath (120 o C) and allowed to react (24 h, 300 RPM, argon sparged). Following the reaction, the product was decanted and allowed to solidify before washing (chloroform, approx. 15 mL). The collected product was dried in vacuo overnight prior to analysis.

Surfactant performance testing Tensiometry
Surface tension (σ) measurements were made using the du Noüy ring method and a bubble tensiometer, measuring equilibrium and dynamic surface tensions, respectively. Equilibrium surface tension measurements were made on a Lauda Tensiometer TD3 using water (60 mL) at a polyester concentration of 0.5% w/v while bubble tensiometer measurements were made on a SITA t100 Bubble Pressure tensiometer with aqueous solutions containing 1% wt/v of polyester. The polyesters, poly(sorbitol adipate) and poly(sorbitol succinate) along with the corresponding laurate and stearate derivatives were all analysed by both methods. The following commercial surfactants; Tween TM 20, Tween TM 28, Tween TM 80, Pluronic TM L35, Pluronic TM L121 and NatraGem TM E145 were also assessed for comparison. All measurements were performed at room temperature (20 °C) and analysis was automated, with repeats made until a standard deviation of ≤ 0.1 mN/m was obtained (≥ 5 measurements). Adapted from Zuidema and Waters. 3

Critical micelle concentration determination
The CMC was determined tensiometrically using an automated Wilhelmy plate. The system contains a fully automated micro dispensing unit which enables a high number of measuring points at a broad concentration range (0.1-10000 mg/L).

Dynamic Light Scattering (DLS) and Z-potential
Particle size analyses were performed by DLS utilizing a Zeta sizer Nano spectrometer (Malvern Instruments Ltd) equipped with a 633 nm laser at a fixed angle of 173°. The same instrument was used to measure the Z-potential of the produced NPs. Analysis was performed at 25 °C on a 1 mL sample (at a concentration ranging from 0.06 -0.50 wt%). All experiments were performed in triplicate on the same sample.

Loading of micelles
Polysorbitol based surfactants (0.1 g) were added to water (1 mL) in order to make a 1 w/v% stock solution. Coumarin 6 (0.1 mL, 1 w/v% in DCM) was added to the stock solutions. DCM was evaporated and the solutions were agitated overnight. The samples were filtered (0.45 m, Millex. L.G, Millipore, USA) and diluted 6-fold in THF before being analysed by UV-Vis spectroscopy.

Dispersion of Hydrophobic dye
Furthermore, all of the solutions were passed through 0.45 µm syringe filters to eliminate undissolved coumarin 6, so it is possible that larger self-assembled compounds have been removed during this step. This could be a concern for surfactants which pack into larger aggregates.  . Spectrum (IV) shows poly(sorbitol adipate) synthesised using a 1:1 molar ratio of reagents and K 2 CO 3 , included as a comparison. Note the very strong similarities between the spectra of the polymers synthesised enzymatically and using K 2 CO 3 , and the increase in higher MW peaks when using equimolar amounts of reagents. The corresponding ion adduct (Na + or K + ) is displayed in the right-hand corner of each spectrum. S * = sorbitan.
Quantitative 13 C-NMR of PSA Figure S3 Quantitative 13 C-NMR spectra in DMSO-d 6 of poly(sorbitol adipate), synthesised at a 1:1 molar ratio of reagents using (I) K 2 CO 3 , with (II) highlighting the area of the spectrum associated with the polyol repeat unit. As a comparison (III) shows the equivalent compound synthesised enzymatically and (IV) shows D-sorbitol (not quantitative). Note the similarity between products synthesised using K 2 CO 3 and Novozym 435, especially peaks at ~ 66 ppm highlighting derivatisation at the primary hydroxyl positions on D-sorbitol.

Solubility testing in scCO 2
Solubility tests in scCO 2 were carried out in a high-pressure fixed volume view cell. This was used to determine the temperature at which the reactants would liquefy or solubilise, as in experiments performed by Bratton et al. 4 The melt temperature of the reagents (95 °C and 150 °C for D-sorbitol and adipic acid respectively, at atmospheric pressure) was not visibly affected by the addition of scCO 2 when heated up to 100 °C as shown in the view cell below. This is highly beneficial when conducting scCO 2 extraction. Lack of solubility in scCO 2 ensures the reagents and product remain in reaction, whilst the scCO 2 soluble condensate is removed. Upon depressurisation significant foaming was observed for both reagents, indicating that scCO 2 can penetrate the reagents (Figure S 4, III). This could be an advantage of employing scCO 2 as it may lead to a reduction in viscosity, as identified previously when processing polyesters such as PCL and PLA. 5,6 Figure S5 1 H-NMR spectra of poly(sorbitol adipate) laurate (PSA-L), in (I) DMSO-d 6 and (II) All D 2 O. Peaks are clearly detectable in DMSO, whilst the lauric acid resonances (red rectangles) are broadened and strongly suppressed in water. This strongly supports the idea that aggregates with a hydrophobic core are formed in aqueous environments. The fraction collected from purifying PSA-L after 6 hours using CHCl 3 has been included as (III) and is identified to be unreacted lauric acid.